322
control protein activity [1]. Importantly, protein interactions are
highly context-dependent. Each interaction partner must be present in threshold amounts and in the right state or conformation,
and this may depend on the presence of a third partner at an allosteric site. As such, a signaling protein’s activity in its natural context can dramatically differ from its activity in vitro or even in a
heterologous expression system. These possible discrepancies can
lead us astray when studying signaling networks as the sum of individual parts rather than as a complex system of functional interdependence. Conversely, changes in network interactions, or
“rewiring,” can alter network output in myriad ways, from simple
changes in signal dynamics to more complex changes in input/
output relationships [2].
Despite their significance, network interactions are often overlooked when studying protein function and evolution; furthermore, they are also under-utilized in protein engineering.
Functional changes during protein evolution are typically studied
by either comparing proteins from related organisms, or by introducing mutations in a gene and observing the consequences of
those changes in the immediate vicinity of the function being analyzed. For example, when studying the evolution of a signaling
kinase, one could analyze how changes in the amino acid sequence
of that kinase affect binding to upstream activators, downstream
substrates, scaffolds, or other regulators. A conceptually different
approach is to investigate how those changes affect the function of
the network, rather than the function of an individual gene. For
that, it is necessary to look beyond the direct interactions between
the protein being analyzed and its close network neighbours. A
particularly powerful approach for studying how changes in an
individual gene affect network function consists in combining random mutagenesis with selection, a procedure known as directed
evolution [3]. To take full advantage of this approach, the method
of mutagenesis, as well as the context of selection, should be carefully considered. As such, we provide here some guidelines for the
directed evolution of signaling proteins in a way that emphasizes
how changes in individual components affect overall network function. We then follow with detailed protocols for mutant library
creation and selections, focusing on how different types of mutagenesis will allow us to explore very different evolutionary trajectories. These are written with the yeast mating pathway in mind, an
example of a highly connected signaling pathway, but they can be
used for other contexts as well.
In any evolutionary process, a source of variability is needed,
and so directed evolution begins with gene mutagenesis. This process has improved considerably since the use of chemical and physical mutagens. Today, these methods have been supplanted by a
wide range of approaches, ranging from error prone PCR [4] , in
which random mutations are introduced in a specific gene, to more
Raphaël B. Di Roberto et al.
control protein activity [1]. Importantly, protein interactions are
highly context-dependent. Each interaction partner must be present in threshold amounts and in the right state or conformation,
and this may depend on the presence of a third partner at an allosteric site. As such, a signaling protein’s activity in its natural context can dramatically differ from its activity in vitro or even in a
heterologous expression system. These possible discrepancies can
lead us astray when studying signaling networks as the sum of individual parts rather than as a complex system of functional interdependence. Conversely, changes in network interactions, or
“rewiring,” can alter network output in myriad ways, from simple
changes in signal dynamics to more complex changes in input/
output relationships [2].
Despite their significance, network interactions are often overlooked when studying protein function and evolution; furthermore, they are also under-utilized in protein engineering.
Functional changes during protein evolution are typically studied
by either comparing proteins from related organisms, or by introducing mutations in a gene and observing the consequences of
those changes in the immediate vicinity of the function being analyzed. For example, when studying the evolution of a signaling
kinase, one could analyze how changes in the amino acid sequence
of that kinase affect binding to upstream activators, downstream
substrates, scaffolds, or other regulators. A conceptually different
approach is to investigate how those changes affect the function of
the network, rather than the function of an individual gene. For
that, it is necessary to look beyond the direct interactions between
the protein being analyzed and its close network neighbours. A
particularly powerful approach for studying how changes in an
individual gene affect network function consists in combining random mutagenesis with selection, a procedure known as directed
evolution [3]. To take full advantage of this approach, the method
of mutagenesis, as well as the context of selection, should be carefully considered. As such, we provide here some guidelines for the
directed evolution of signaling proteins in a way that emphasizes
how changes in individual components affect overall network function. We then follow with detailed protocols for mutant library
creation and selections, focusing on how different types of mutagenesis will allow us to explore very different evolutionary trajectories. These are written with the yeast mating pathway in mind, an
example of a highly connected signaling pathway, but they can be
used for other contexts as well.
In any evolutionary process, a source of variability is needed,
and so directed evolution begins with gene mutagenesis. This process has improved considerably since the use of chemical and physical mutagens. Today, these methods have been supplanted by a
wide range of approaches, ranging from error prone PCR [4] , in
which random mutations are introduced in a specific gene, to more
Raphaël B. Di Roberto et al.